Pseudopotentials and basis sets database

How can I get a pseudopotential and basis set for the elements of my system? Is there a database or do I have to generate the pseudopotentials by myself and design the basis sets?

A database of LDA and GGA-PBE .psf pseudopotentials (basis sets) is available here however, as you can find written in the web page:

Use at your own risk. Neither the Siesta nor Abinit teams are responsible for the suitability of any pseudo for your particular problem. The Siesta team does not test the pseudos. Before their usage in real world challenges pseudopotentials and basis sets should be tested in very well-known situations.

As alternative you can generate the pseudopotential using the code ATOM.

SIMUNE, to overcome this difficulty, is developing a database of transferable pseudo and basis set, ready to be used in a SIESTA calculation, please visit the database section at .

Moreover, notice that SIESTA users that compiled the (soon to be merged) version of SIESTA with support for the new PSML format will be able to use the pseudopotentials available at

More info can be found here

Does the latest version Siesta support PSML format pseudopotential?
I want to do some calculation on La atom but I could not find corresponding La.psf file from here
I try to generate an La.psf by myself with ATOM, my input file was
pe Lanthanum Guess Ecut ~ 40Ry l=0 as local
n=La c=car
0.0 0.0 0.0 0.0 0.0 0.0
11 4
6 0 2.00 0.00
6 1 0.00 0.00
5 2 1.00 0.00
4 3 0.00 0.00
3.50000 4.10000 3.50000 3.50000
#23456789012345678901234567890123456789012345678901234567890 Ruler
and I succeed to get the pseudopotential, but when I do calculation on La, I got some warning in result file

GHOST: No ghost state for L =  0
KBproj: WARNING: Rc( 0)=      5.1138
KBproj: WARNING: Cut of radius for the KB projector too big
KBproj: WARNING: Increasing the tolerance parameter eps
KBproj: WARNING: might be a good idea

Will this warning be malignant to my results?

Thank you advance for reply!

You should decrease the cutoff radius in your pseudo generation input. Try to lower the p channel cutoff radius, the warning message should disappear.
Anyway, that does not guarantee that the pseudo will be appropriate, we suggest to test it using the Atom testing tool and on after by running SIESTA on a known system.

Moreover, I would suggest to try to include the 5s and 5p orbitals in valence for the generation of the pseudopotential, as it is done in the psml pseudopotential that you can find for La in However, this might require the inclusion of a specific basis set block in the fdf input file.

SIESTA 4.1 supporting psml can be found in trunk-psml or MaX branches.